How can Tik Tok avoid being the internet’s shiny new toy.

Many in the social media space have noted that Tik Tok has a user retention problem. According to Business Of Apps, Tik Tok’s 7-day retention in January 2020 is 26%. Facebook’s CEO Mark Zuckerberg said in July 2019 that “what we’ve found is that their [TikTok’s] retention is actually not that strong after they stop advertising”. Considering that Bytedance spent around $1 Billion advertising Tik Tok in 2018, the inability to retain users not only burns the pockets of investors greatly, it is also a sign of a much deeper problem in the long run. As is widely acknowledged in the product world, growing a product that lacks retention is like filling a leaking tank with drilled hose: you can pump as much water as you can, but it will come at an extremely high cost. If Tik Tok continues to churn through users without fixing its funnel, it could well end up like Twitter, an app that is universally known amongst all, but one that most abandon after a while.

Just what exactly is the nature of Tik Tok’s retention problem? If we were to take the figures provided by Business Of Apps, Tik Tok’s short run retention is 26%. Whilst this is not half as good as Facebook’s 45% and Snapchat’s 32%, a different set of statistics by App Annie suggests that it is in fact 39% in September 2020, which shows that it is on the rise. It is hard to determine which set of figures are truly accurate. But if we assume that Tik Tok’s short run retention is indeed improving, then Tik Tok must have done a good job at the following things since June 2019: providing new users solid experience, getting them to experience core value as quickly as possible by recommending top quality curated content, and having a good onboarding UX. My own experiences with the app suggests that there is nothing strictly wrong on the new user experience front.

However, if we were to look at long run retention, we find a less than promising picture. According to one source, Tik Tok’s long run retention (>30 days) is around 4%. It is far lower than Instagram’s 10% and Snapchat’s 8%. Many other sources, including Facebook’s CEO, also suggest that Tik Tok has a long term retention problem. Why might that be the case?

One explanation, provided by Charit Narayanan in his Medium article, is that Tik Tok is a “niche app” that simply does not have mainstream appeal, unlike Facebook or Instagram. He argues that Tik Tok’s slightly odd, high-energy, meme like culture attracts teenagers as much as it alienates other demographics. This argument might be true: after all, Tik Tok’s initial expansion strategy, as well as platform design, was tailored almost exclusively for the teenage market. But a closer look at the statistics suggests otherwise, at least for some parts of the world. In places like China and India, for instance, more than 60% of Tik Tok’s users are between the 25–44 age bracket. Tik Tok therefore has the potential to capture a more mainstream appeal in many parts of the world, but doing so would typically require an active effort in maintaining a content ecosystem of a different kind. (Read my article on the stark differences between Douyin and Tik Tok).

The second reason, I would suggest, for Tik Tok’s poor long run retention rates lies in its recommendation system. Tik Tok’s recommendation engine is a blunt tool that, whilst offering content that is interesting enough, stops short of providing content that people truly want in the long run. According to a study done by Sean Wang, 48% of the people who stopped using Tik Tok was due to the lack of variety of content. A common complaint that I hear around me is also that Tik Tok’s content suffers greatly from homogeneity. Of course, this is not to say that there isn’t plenty of interesting content on Tik Tok. It is just that it is hard to get personalization right — what you think I like might not be what I really like, what I like for now might not be what I will continue to like in days to come, and what I like on a surface level to pass my time might not be what I truly value or need in any deep way. The truly successful apps, as I will explore later, creates hooks based on deep desires and needs.

There is also a third reason that we should probably consider. Previously, I have argued that Tik Tok is primarily a performative social entertainment site that contrasts with the life-logging models of Facebook, Snapchat and Instagram. I’ll reproduce the diagram used in my explanations below:

One particular characteristic of this kind of app is that it has moved away from the importance of friend networks. Tik Tok is more a stage where you perform for strangers than a tool for connecting with close friends, and it is so by design. There is one important trade off for this kind of system design. When you deprioritize friend networks, the burden of retention will fall greater on the shoulders of content. The reason why Facebook and Instagram are so successful in retaining users is because they know that human beings are social creatures that care primarily about the content of their friends. Users return because it fulfils a fundamental desire in them: the desire for connection and intimacy. When you shift away from the lifelog model, you are losing the ability to tap on our most primal calls to return, and shifting all these weight onto the shoulders of content.

Why? Think about it this way. The reason people return to a social media like Snapchat, for instance, is because they derive value from the intimacy experienced when they receive a snap. If intimacy is completely missing from the equation, then content is the only thing anchoring people’s willingness to return. Once content becomes the central piece on which retention relies upon, the probabilities of user churn is greater because there are just way too many ways for content to suffer mismatch with user preferences. And when the supply of quality content runs low due to increase in usage, content becomes a limiter to growth and retention.Within this formula, growth and selection of quality content has to be on par with, or faster than, the usage levels of a majority of users for Tik Tok’s ecosystem to be sustainable — and for users to be retained.

To resolve this issue, a there are a few things Tik Tok can consider. First, maintaining and developing a pool of high quality content that caters to a wider group of users is key. However, this is really easier said than done. Quality user generated content is hard to come by in the first place, and in many ways it is not something that the platform has direct control over. In addition, “Quality content” is not one coherent pool that Tik Tok can draw upon and distribute to all users. When content is optimized according to individual users’ value systems, latent desires, preferences, and other traits, the pool of quality content becomes even smaller. Consequently Tik Tok’s content team has to shoulder heavy burdens of ensuring quality content experience in hundreds of demographic verticals in order to best retain all of them.

If relying on users to consistently produce world-class content is too risky a retention strategy, then Tik Tok should consider unearthing hidden value by better “content to value-system” matching and optimization through user insights. Every user sees the world according to their own normative assumptions and from the perspective of their own “cultural lens”. Ethnographic studies into the value systems and preferences of differing sub-groups, according to race, ethnicity, gender, “liked” videos, and even time of day when using Tik Tok, will probably reveal insights that can be fed into the recommendation engine to capture the most value across the biggest sub-group of users.

Finally, if Tik Tok wants to switch things up a notch, it should probably consider integrating life-log features into the platform. By that, I don’t mean to suggest that Tik Tok should change its direction completely and become an app for recording one’s life like Facebook and Instagram. Rather, there are other small things it can do to shift it towards that direction in order to reap some of its benefits. For instance, it could take a more proactive approach to prompt users to invite their friends onto Tik Tok, or to suggest friends to follow. It could tweak its culture to prioritize content by close friends more, via its For You page, once in a while. It could optimize Users’ personal profile page to show which close friends have followed a particular account. It could optimize its buttons to show which close friends have liked the videos. It could even add a feed that displays the videos that their friends reposted, which Dou Yin has recently done. These are all small things that Tik Tok can do to emphasize friendship networks a little bit more. When done properly, Tik Tok product managers would soon realize that users are not returning to Tik Tok simply because they are bored(as is revealed in Tencents’ study), but because they find that it fulfils their desires for intimacy, connection, and a sense of communal belonging. All these are deep desires. Once captured, the pressure of retention will be alleviated from content, providing room for greater content diversification, as well as adding additional fuel for organic growth.

Retention is ultimately a question of matching people’s latent desires to the functionalities or content provided by the app. It is a complex question that requires a lot of trial and error (read, A/B testing and cohort analysis*) to get right. But when Tik Tok loses its shine and people grow bored with the lack of variety in formats, it is the relationships and friend networks that will cause them to keep returning.

**************

*For an example of how that can be done, check out a related article of mine: “A full-stack ethnographic approach to User insights”